 |
 |
|
|
Alternative Review of the Fourth Week's ResultsIn our usual analysis we describe leading ten Expert Advisors. This time we decided to change our approach and prepared the analysis of absolutely all Expert Advisors participating in the Automated Trading Championship 2008. Last week we tried to estimate results of the top ten on the basis of normalized trades. This time we recalculated trading results of all Participants: all profits and losses of executed trades were recalculated supposing EAs traded 0.1 lots per a trade. As it was stated earlier, this is done for a complete elimination of the influence of risk and money management systems. After recalculation we obtained each Participant's Balance and Equity values based on normalized trades. After that we made up a mini-rating of Participants according to obtained Equity values to see who would be in the top ten by the results of normalized trades. The new table has almost nothing in common with the real top ten of Participants. Only one leading Participant is included into the new table – Gorez.
Gorez is on the first place in this table based on recalculated values with Equity equal to $20,219. It is explained by the fact that Gorez always opened positions of the same size - 1 lot, while other top ten Participants opened positions of larger volumes. According to Balance value the first place would be occupied by solandr whose fixed profit on normalized trades would be $15 517. One more interesting fact is that the alternative top ten containы three multicurrency Expert Advisors that trade two or more symbols: Participants notused, Hendrick and solandr. This is really unexpected, because multicurrency EAs rarely appear in our reports.
Besides, we see that all one-currency EAs owe their success to the pair GBPJPY. This currency pair has fallen since the Championship start almost by 5,000 points, 2,000 of which were passed on Friday, October 24th, 2008. The only exception is the Participant kotegh who trades EURUSD and is on the 8th place of our mini-rating. We can conclude that success often depends on a correctly chosen symbol or market. Winners of previous Championships did not trade such a "bulky" pair as GBPJPY. Having risen to the top one can easily fall the same quickly. We will soon see, whether four week's leaders can stay on their current positions. Created: 2008.10.31 Author: MetaQuotes
Pages:
2 I think you are both right. These numbers shown here (when you fix the lot size for any bet to whatever number, 0.1 lots, or 1 lot, or 20 lots) then what you get is the amount of pips won for any bet. You can find this very easy, by subtracting the open price from the close price for long trades, and viceversa for short trades. Then what is left, you just have to multiply with the amount of lots. In this respect, is normal that experts trading GJ and being profitable will show up in the top, because GJ is moving more then all other pairs at this time. This is just a math consequence, the pairs with a price close to 0 move slower then the pairs far away. For example, a (fictive) pair USDGBP would "cost" now 0.6133, and if this (fictive) pair moves 5 pips to 0.6138, then the (real) pair GBPUSD moved in fact from 1.6305 to 1.6292, that is 13 pips! This was just an (fictive) example to show that pairs close to 0 move slower. Pure math. You can put any (real) pair instead of USDGBP, the effect is the same, only the correlation is more complex. I selected especially a very simple correlation, for the sake of example, as USDGBP=1/GBPUSD, but the idea is the same if you select EURGBP and compare it with GBPJPY and EURJPY, they are correlate through a FPI ring: EURGBP=EURJPY/GBPJPY. The math is exactly the same. So, this calculus shown in the article is just a find out of "which pair move more", that is of course GBPJPY, having the highest price (and as a consequence - highest volatility) now. We could call the article "which competitor made most pips", because the amount of lots, as I said, is just a constant. If you consider 20 lots, you will get EXATLY the same table, same competitors, in the same order, but with a bigger equity for each player in the list. So, to make the calculus more realistic, we should transform the number of pips made by every player into "EURUSD-pips", by dividing or multiplying it within the FPI ring with USD and the pair traded. For example if the payer traded GBPJPY, then the FPI ring is EURUSD-GBPUSD-GBPJPY-EURJPY, and we have to multiply two times and divide once. This way, we will take the "volatility" out of the equation, and get a REAL estimation of the "how many pips the players should have been won if they all should have traded EURUSD" (sorry for the grammar :D) That would be really interesting, and more accurate :D Who is interested, he can make his own calculus, with hust a simple script. I am still missing Phy... :P jlpi wrote: taking a same fixed lot size for all pairs is a wrong approach as it is proven by the fact that you get almost only experts with GBPJPY on the top of the list. Because of course a same lot size will favorize the more volatile pairs. EURGBP experts wouldn't stand a chance for instance. It very strange that all Participants don't trade same pair GBPJPY. May be only you know this fact that GBPJPY is very volatile :) jlpi wrote: Think that when you trade for instance with a SL a fixed percentage of your balance, usually a SL in EURGBP would be closer in pips than a SL in GBPJPY and then to get the same risk you will have a lot size smaller for GBPJPY than for EUGBP. Thank you for reveal this knowledge. jlpi wrote: If you want to put a fixed lot size you should make different groups for EAs on each currency (then would be interesting to see what you do with the multicurrency ones) It wouldl be another story. jlpi wrote:
Before publishing that you could have thought few seconds when you get such kind of results, what was the reason behind it. Did you even thought that a lot size inversely proportional to the volatility of the pair would have been more reasonable. I guess that you could find few programmers (or at least one) iin MetaQuotes able to make the code to run that on all experts. You can calculate these calculations yourself with MQL4 script as I did it. 2008.10.31 23:06
taking a same fixed lot size for all pairs is a wrong approach as it is proven by the fact that you get almost only experts with GBPJPY on the top of the list. Because of course a same lot size will favorize the more volatile pairs. EURGBP experts wouldn't stand a chance for instance. Think that when you trade for instance with a SL a fixed percentage of your balance, usually a SL in EURGBP would be closer in pips than a SL in GBPJPY and then to get the same risk you will have a lot size smaller for GBPJPY than for EUGBP. If you want to put a fixed lot size you should make different groups for EAs on each currency (then would be interesting to see what you do with the multicurrency ones) So what this study shows is that GBPJPY had the greater volatility. What a discovery ! Before publishing that you could have thought few seconds when you get such kind of results, what was the reason behind it. Did you even thought that a lot size inversely proportional to the volatility of the pair would have been more reasonable. I guess that you could find few programmers (or at least one) iin MetaQuotes able to make the code to run that on all experts. 2008.10.31 19:53
|
Sponsors
Media Sponsor
Organizer
|
Previous
I think you are both right. These numbers shown here (when you fix the lot size for any bet to whatever number, 0.1 lots, or 1 lot, or 20 lots) then what you get is the amount of pips won for any bet. You can find this very easy, by subtracting the open price from the close price for long trades, and viceversa for short trades. Then what is left, you just have to multiply with the amount of lots.
In this respect, is normal that experts trading GJ and being profitable will show up in the top, because GJ is moving more then all other pairs at this time. This is just a math consequence, the pairs with a price close to 0 move slower then the pairs far away. For example, a (fictive) pair USDGBP would "cost" now 0.6133, and if this (fictive) pair moves 5 pips to 0.6138, then the (real) pair GBPUSD moved in fact from 1.6305 to 1.6292, that is 13 pips! This was just an (fictive) example to show that pairs close to 0 move slower. Pure math. You can put any (real) pair instead of USDGBP, the effect is the same, only the correlation is more complex. I selected especially a very simple correlation, for the sake of example, as USDGBP=1/GBPUSD, but the idea is the same if you select EURGBP and compare it with GBPJPY and EURJPY, they are correlate through a FPI ring: EURGBP=EURJPY/GBPJPY. The math is exactly the same.
You also can view GBPJPY as JPYGBP. We have GBPJPY for 190.00 - 139.00 ~ 5100 pips in one hand. But on other hand we have for GBPJPY 1/190 - 1/139 = 0.005263 - 0.007194 ~ 2000 pips.